3D environments provide new possibilities for musical interaction. They allow musicians to manipulate and visualize large sets of sound processes associated to 3D objects by connecting graphical parameters to sound parameters. Several of these audiovisual mappings can be combined on a single 3D object. However, this brings up the issues of the choice of these mappings and of their combinations. We conducted a user study on sixteen musicians to evaluate audiovisual mappings and their combinations in the context of 3D musical interaction. This user study is composed of three experiments. The first experiment investigates subjects preferences for mappings between four perceptual sound parameters (amplitude, pitch, spectral centroid and noisiness) and ten graphical parameters, some of them specific to 3D environments. The second experiment focuses on efficiency of single mappings in an audiovisual identification task. The results show almost no significant differences, but some tendencies, which may indicate that the choice of mappings scales is more important than the choice of the mappings themselves. The third experiment investigates the efficiency of mappings combinations. The results indicate no significant differences, which suggest that it may be possible to combine up to four audiovisual mappings on a single graphical object without any performance loss for musicians, if they do not disrupt each other.